Designer solvents for the extraction of glycols and alcohols from aqueous streams

The separation of polar compounds from aqueous streams is one of the most energy intensive operations within the chemical industry, because of the formation of hydrogen bonds that should be broken and the high heat of vaporization of water. Important bulk chemicals like glycols and alcohols produced from petrochemical feedstock or renewable sources through fermentation processes are classified in this category. In this thesis, the recovery of low molecular weight diols, such as Mono Ethylene Glycol (MEG), 1,2-Propanediol also known as propylene Glycol (PG) and 2,3-Butanediol (BD) that are particularly difficult to separate due to their high water affinity and important alcohols like 1-Butanol (BuOH) were studied. Their recovery by conventional multiple effect distillation is associated with high energy consumption; therefore liquid-liquid extraction technology can be a promising alternative since it can be more energy efficient. However, for conventional solvents the distribution coefficients are generally insufficient to achieve efficient extraction at low concentrations as encountered in the chemical synthesis of diols or production through fermentation. For this reason, the use of novel extraction solvents like reactive extractants or ionic liquids is needed to improve the glycol distribution coefficient and selectivity. Boronic acids derivatives were studied as they are known because of their good ability to form complexes with cis-diols. Naphtalene-2-Boronic Acid (NBA) was selected as extractant and it was diluted in 1-Ethyl-Hexanol and octanol. Aliquat 336 (N-Methyl-N,N-dioctyloctan-1-ammonium chloride) was applied as counterion to facilitate the complexation between NBA and MEG. 1-Ethyl-Hexanol was the better diluent. The partition coefficient of MEG in 1-ethyl-hexanol was 0.0025, and distributions and selectivities up to 0.026 and 0.089, respectively (at pH 11, 298K) were observed with NBA and Aliquat 336 in equal amounts at 0.2 mol/L. This maximum distribution is around 10 times better than the conventional solvents. Nevertheless further improvements in the distribution and selectivity towards the glycol are required, which could be provided by advanced solvents like ionic liquids. In Chapter 2, experimental work and molecular modelling simulation with COSMO-RS were used to support the solvent screening for (MEG). The ionic liquid design and tailoring to optimize the glycol distribution coefficient (D) and selectivity (S) was done by employing the sigma electron profile. As a result, the glycol distribution coefficients improved compare to the conventional solvents when combining a tetraoctyl ammonium cation of the IL with a carboxylate, phosphinate and boronate anion. These ILs were tailor made and evaluated in LLE experiments. They outperformed the other solvents tested with DMEG up to 0.45, and S up to 3.2 vs DMEG = 0.04 and S = 0.95 for 2-ethyl-hexanol for initial feed concentrations of 20% wt MEG. For the best performing ionic liquid tetraoctyl ammonium 2-methyl-1-napthoate [TOA MNaph], liquid-liquid equilibrium data were determined and the NRTL and UNIQUAC thermodynamic models were constructed for the three different glycols. The results, presented in Chapter 4, show that both models can properly describe the experimental data. These thermodynamic models were used to develop conceptual process designs in Aspen Plus ® and compared the different processes for the production of MEG and PG with two different technologies, conventional triple effect distillation (MED) and solvent extraction (LLE) using [TOA MNaph].The results showed that the LLE alternative could provide energy savings >50% compared to the current state-of-the-art three effect distillation technology (94% for MEG 20% wt from the petrochemical process and 54% for PG 10%wt from a fermentation process). Regarding CAPEX, the conventional technology is always preferable because less equipment is required, while for the LLE technology the CAPEX is higher due to the solvent cost, the equipment in solvent recovery section and the additional heat exchangers required for the heat integration in the process. The purification of PG has the lowest CAPEX because a lower solvent to feed ratio is required compared to MEG extraction. According to a total annualized cost analysis at the current crude oil prices, the purification of PG from a fermentation broth via LLE could be an advantageous technology to replace MED. For the MEG production we can say that currently the LLE process is not a suitable option and that a significant increase in crude oil prices should occur before the use of LLE technology with this IL can become feasible. In Chapter 6, liquid-liquid extraction of butanol from water, employing [TOA MNaph] was evaluated against distillation and extraction with conventional solvents. The results show that this IL yields the best distribution coefficient and very high selectivity (DBuOH=21, S=274), compared to the benchmark solvent oleyl alcohol (DBuOH=3.42, S=192). The conceptual design study showed that butanol extraction with [TOA MNaph] requires 73% less energy than in conventional distillation (5.65 MJ/kg BuOH vs 21.3 MJ/kg for distillation). Finally it was concluded, that the feasibility of the LLE with ILs is strongly dependent on the glycol distribution coefficient and selectivity achieved. As the conceptual process design and economic evaluation showed, there are still challenges and improvements to extend the search for additional solvents (not limited to ionic liquids), especially for extremely polar compounds like MEG, which requires an even higher capacity and selectivity than could be achieved in this study